The NVIDIA Titan V Preview - Titanomachy: War of the Titans
by Ryan Smith & Nate Oh on December 20, 2017 11:30 AM ESTGaming Performance
Sure, compute is useful. But be honest: you came here for the 4K gaming benchmarks, right?
Already after Battlefield 1 (DX11) and Ashes (DX12), we can see that Titan V is not a monster gaming card, though it still is faster than Titan Xp. This is not unexpected, as Titan V's focus is quite far away from gaming as opposed to the focus of the previous Titan cards.
Despite being generally ahead of Titan Xp, it's clear Titan V is suffering from lack of gaming optimization. And for that matter, the launch drivers definitely have bugs in them as far as gaming is concerned. Titan V on Deus Ex resulted in small black box artifacts during the benchmark; Ghost Recon Wildlands experienced sporadic but persistant hitching, and Ashes occasionally suffered from fullscreen flickering.
And despite the impressive 3-digit FPS in the Vulkan-powered DOOM, the card actually falls behind Titan Xp in 99th percentile framerates. For such high average framerates, even a 67fps 99th percentile can reduce perceived smoothness. Meanwhile, running Titan V under DX12 for Deus Ex and Total War: Warhammer resulted in less performance. But with immature gaming drivers, it is too early to say if these are representative of low-level API performance on Volta itself.
Overall, the Titan V averages out to around 15% faster than the Titan Xp, excluding 99th percentiles, but with the aforementioned caveats. Titan V's high average FPS in DOOM and Deus Ex are somewhat marred by stagnant 99th percentiles and minor but noticable artifacting, respectively.
So as a pure gaming card, our preview results indicate that this would not the best gaming purchase at $3000. Typically, a $1800 premium for around 10 - 20% faster gaming over the Titan Xp wouldn't be enticing, but it seems there are always some who insist.
111 Comments
View All Comments
praktik - Wednesday, December 20, 2017 - link
Actually probably both XP and V could run 4k Crysis pretty well - do we need 4xssaa @ 4k??Ryan Smith - Wednesday, December 20, 2017 - link
"do we need 4xssaa"If it were up to me, the answer to that would always be yes. Jaggies suck.
tipoo - Wednesday, December 20, 2017 - link
Do they plan on exposing fast FP16 in software? When consumer Volta launches maybe?Ryan Smith - Wednesday, December 20, 2017 - link
Nothing has been announced at this time.Keldor314 - Wednesday, December 20, 2017 - link
The part of the article about Volta no longer having a superscalar architecture is incorrect. Although there is only one warp scheduler per SM partition (what do you call those things anyway?), each clock cycles only serves half a warp, so it takes two clock cycles for an instruction to feed into one of the execution pipelines, but during the second cycle, the warp schedular is free is issue a second instruction to one of the other pipelines. IIRC, Fermi did this too.mode_13h - Wednesday, December 27, 2017 - link
Also, the part about per-thread PC and Stack is misleading. Warps are still executing (or not executing) from a single instruction sequence. The threads within a warp are not concurrently executing different instructions, nor are threads being dynamically shuffled between different warps - at least, not at a hardware level.MrSpadge - Wednesday, December 20, 2017 - link
> Sure, compute is useful. But be honest: you came here for the 4K gaming benchmarks, right?Actually, no: I came for compute, power and voltage.
jabbadap - Wednesday, December 20, 2017 - link
Interesting, so it have full floating point compute capabilities 1*fp64 -> 2*fp32 -> 4*fp16 + Tensor cores. But that half precision is only for CUDA? So no direct3d 12 minimum floating point precision.Native7i - Wednesday, December 20, 2017 - link
So it looks like V series focused on machine learning and development.Maybe rumors are correct about Ampere replacing Pascal...
extide - Saturday, December 23, 2017 - link
Maybe, I mean GP100 was very different than GP102 on down, so they could do the same thing..